Modal
DIY AI Tests Three Video Models in a 9-Clip Product Trial
DIY AI has published nine product-video tests comparing Veo 3.1 Lite, Runway Gen-4 Turbo and Kling 3.0 in its Studio platform. The publisher recommends Veo 3.1 Lite as…
BusinessModal lifts valuation to $15.75 billion and passes $1 billion run rate
Modal has raised its valuation to $15.75 billion and says it has crossed $1 billion in annualized run rate. The company builds AI building blocks that let developers…
Modal releases Spindle, an open-source Tinker-compatible training server
Modal has released Spindle, an open-source training server that is compatible with the Tinker API from Thinking Machines and runs inside a customer's own Modal…
Modal launches Dojo, a library for production-grade LLM training
Modal has released Modal Dojo, a library that gives teams building blocks, tuned defaults and observability for training LLMs, including with reinforcement learning.…
Modal launches Sticky Sessions to keep users on the same container
Modal has launched Sticky Sessions, a feature that sends every request in a session to the same container while keeping serverless autoscaling. It is built on the…
Modal Says NVIDIA Vera CPUs Can Double Sandbox Throughput in Early Tests
Modal says early benchmarks of NVIDIA's Vera CPU show up to double the task throughput of an equivalent x86 GNR machine, without slower individual tasks. The company…
Modal swaps virtio-fs for block devices to speed up VM Sandbox filesystems
Modal has rebuilt the filesystem behind its new VM Sandboxes after finding that virtio-fs made everyday operations far slower than on its container platform. According…
Meta's Muse Spark Open Weights Now Available on Modal Auto Endpoints
Meta has released weights for Muse Spark Open, a model built for agentic coding and long-horizon tasks, and it is now available on Modal Auto Endpoints. Modal says the…
Modal tops $1B annualized run rate and raises $750M Series D
AI infrastructure company Modal says it has crossed $1B in annualized run rate and raised a $750M Series D at a $15.75B post-money valuation. Accel led the round…
Modal shows stable RL training with mixed-precision, mixed-accelerator samplers
Modal says it has trained a large model with reinforcement learning (RL) using samplers on six kinds of accelerators and several numeric precisions, matching a…
Modal launches Endpoint Candidates to test inference changes on live traffic
Modal has made Endpoint Candidates generally available for all workspaces. The feature lets teams stage alternative deployments of a Dedicated Endpoint, mirror a sample…
TypeSafe AI's Jev model passes one trillion tokens in three days on Modal
TypeSafe AI's first model, Jev, scaled to more than one trillion tokens in three days after its public launch, running on Modal's serverless infrastructure, according to…
Modal Clusters reach general availability with one-line multi-node GPU access
Modal has made Modal Clusters generally available, letting developers run multi-node GPU workloads through a single decorator, @modal.clustered. According to the…
Modal launches Sidecars to split trusted and untrusted code in Sandboxes
Modal has introduced Sidecars, isolated containers that run beside a main Sandbox on the same host and form a security boundary between trusted and untrusted code.…
Modal makes VM Sandboxes generally available for AI agents
Modal has made VM Sandboxes generally available, giving AI agents a full Linux virtual machine that can be switched on with one flag. According to Modal, the new runtime…